DTIC · 2020. 2. 20. · AD-A252 880 Edward L. Ginzton Laboratory Stanford University Stanford, California 94305-4085 A Final Report DTIC For ELECTE S JUL 16 1992 D An wA All-Solid-State

AD-A252 880

Edward L. Ginzton LaboratoryStanford University

Stanford, California 94305-4085

AFinal Report

DTIC For

ELECTEJUL 16 1992 AnS D All-Solid-StatewA Chirped Source

forCoherent Optical Radar

ONR Contract Number N00014-88-K-0701

Principal InvestigatorRobert L. Byer, Professor of Applied Physics

Dean of ResearchStanford University

Stanford, California 94305-4085T__ (415) 723-0226

; Lto Jl i r l: A A,':c-li

March 15, 1992

92-0997892 4 20 o 11 IIJIIHII h IUIEH

Abstract

A low-noise, high-power, single-frequency, solid-state laser harmonically con-verted to the green is required to pump a single-frequency, singly-resonant,chirped, optical parametric oscillator. As work toward this new frequencyagile source we have built and injection locked an 18 watt Nd:YAG slavelaser with a 40 mW master laser to produce single frequency operation, gen-erated 6.5 watts of 532 nm radiation at 36% efficiency by resonant secondharmonic generation, and measured spectral and spatial mode characteristicsof this laser. Toward an all-solid-state version of this laser we have designeda diode-laser-pumped Nd:YAG laser. All subsystems of this laser have beenbuilt and tested and the final construction of the laser is currently underway.In addition, we have built both a pulsed singly resonant optical paramet-ric oscillator, and a low threshold cw doubly resonant optical parametricoscillator that operated at 80% conversion efficiency in a single axial mode.

AccedIon ,our t

By.........DO~t ibj tio. I

AvdIldbiiy ' . , !

Statement A per telecon Mathew White I --- I ......ONR/Code 126 Dit Avil a, OctArlington, VA 22217-5000 p

NWW 7/15/92 A-

IntroductionRecent progress under the Office of Naval Research contract ONR N00014-88-K-0701, "An All-Solid-State Chirped Source for Coherent Optical Radar"is described here and in the publications and dissertations resulting fromthis program listed below and contained in the Appendix. We report sevenprinciple accomplishments: 1) demonstration of an 18-watt, single-frequency,injection-locked, TEM00 , Nd:YAG laser, [1] 2) measurement of the phase fi-delity of this regenerative amplifier and the frequency, and intensity noiseof the injection locked laser, 3) conversion of the output of this laser to6.5 watts of single-frequency, TEMoo, green radiation at 532 nm by reso-nant second harmonic generation, [2], 4) the of measurement of the spatialmode characteristics of this lasers [21 and its second harmonic, 5) progresstoward a 13-watt, single-frequency, diode-laser-pumped, solid-state laser, 6)the construction and testing of a pulsed parametric oscillator,[3] and 7) theconstruction of a cw doubly resonant parametric oscillator and the use ofthis 80% efficient device for the generation of squeezed states.[4] This reportis divided into eleven sections. Seven sections describe each of the accom-plishments listed above, two sections list the publications and dissertationsresulting from this program, a section listing the personnel supported and aconclusion.

18-W injection locked lamp-pumped Antares laserThe single-frequency, singly-resonant, parametric oscillator, pump laser re-quirements include: low frequency noise, low amplitude noise, good spatialmode quality and high output power. The approach we have selected tomeet these requirements is to use injection locking where the stability of alow power master oscillator controls the spectral characteristics of the highpower slave oscillator. The low power master is extremely stable and isdescribed in reference [5] which describes frequency stabilization of diode-laser-pumped solid state lasers. The frequency noise of the low power masteroscillator is impressed on the high power slave oscillator by injection locking.The slave laser acts like a regenerative amplifier and reproduces the spectralproperties of the master at high power.[6]

Injection lockingIn injection locking, the phase of a high power oscillator, the slave, is locked tothe phase of a low power oscillator, the master. This preserves the linewidthcharacteristics of the master in the high power slave output. The master

oscillator in our experiment is a monolithic, single axial mode nonplanarring oscillator pumped by a diode laser.[7]

In 1984 Kane and Byer invented the monolithic nonplanar ring oscillator(NPRO) that has several significant advantages over previously demonstratedsolid-state lasers. The nonplanar ring oscillator combines the low noise of acompact monolithic design with the ability to operate in a single directionbecause of the built-in optical diode. To date NPRO's have operated atkilohertz free-running linewidths with noise spectral densities of 20Hz/V'H-zat 1 kHz and 100Hz/Vl/H- at 100 Hz. This spectral density of frequencynoise is three orders of magnitude below that of an Argon ion laser. Thejitter linewidth of this master oscillator is typically less than 10 kHz.[8]

The lamp pumped slave laser uses a ring resonator with the Nd:YAGhead from a Coherent Inc. Antares Model 76-s laser as the gain medium.The ring cavity consists of four flat mirrors and cavity stability is obtainedby the thermal focussing of the Nd:YAG rod. The cavity length is 133 cmwhich corresponds to a free spectral range of 225 MHz. The output couplerhas transmissions, T, = 0.17 and Tp = 0.45, and 18 W of output power wasobtained in a single axial mode when injection locked.

Without the master laser the slave oscillated simultaneously in both direc-tions in 10 axial modes and with 9 watts of average power in each direction.By injection locking the slave with the output of a 40 mW master laser, theslave could be made to oscillate unidirectionally, in a single linearly polarized,TEM00 axial mode at the frequency of the master. When injection locked,the power in the direction of the injected light doubles to 18 W while thatin the opposite direction was extinguished.

A schematic of the laser system is shown in Figure , 1. The master laseris mode matched into the slave laser with a lens and is protected from theslave power, in the event that the slave looses lock and oscillates bidirection-ally, by two Faraday optical isolators. Injection locking is accomplished byPound-Drever [9] locking the slave cavity to the master frequency. The slavefrequency actuation is achieved using two mirrors mounted on PZT pushers.One PZT has high dynamic range but low bandwidth and the other has highbandwidth and low dynamic range. The servo is a cascaded integrator, splitinto fast and slow loops, and provides 56 dB of gain at DC and has a unitygain point of 30 KHz.

The full width of the locking range [1] is given by equation 0.1 where Tis the transmittance of the slave oscillator output coupler, VFSR is the slave

2

FAST PZT.HR SLOW PZT-HA

?dd:YAG -\ F

RF

A~oFR

Figure 1: Injection-locked Nd:YAG Lamp pumped Antares laser system.The FM sidebands are impressed on the single-frequency, diode-laser-pumpedmaster laser. The slave laser cavity is held at the lock point by feeding backthe Pound-Drever error signal through the split servo loops to the two PZTmounted-mirrors. HR's, high reflectors; OC, output couplers; BS, beamsplitter; LPF, low-pass filter.

oscillator free spectral range, and t/ is the spatial mode coupling efficiencyfactor.

AfLock = L/Fs Pmse ( 1)

7T:7T

We measured the locking range by scanning the slave oscillator cavitylength and measuring the width of the frequency discriminant from max-imum to minimum of the dispersive-shaped signal. Figure 2 shows thelocking bandwidth as a function of the square root of the power ratio of the

master to slave. The slope of the line is 13.8 MHz which shows good agree-ment with the calculated value of 16.8 MHz based on r.= 1. The discrepancycan be accounted for by the imperfect spatial and polarization mode overlap.We achieved injection locking with slave powers of up to 18 W with a masterpower of 40 mW and with about 80 percent of the power in the master carrier

for a power enhancement ratio of 450.

3

1.4

.1.2'2

a10U

i slop.- / 1 3.8 MHz0.8

O 04 0.05 0.06 0.07 0.08 0.09 0.10

P/ miate, Slave

Figure 2: Injection locking range versus the square root of the ratio of themaster oscillator power to the slave oscillator power. The data is fit to astraight line with a slope of 13.8 MHz.

Noise propertiesThere are two types of laser noise that will degrade the single-axial-modeparametric oscillator performance: frequency, and intensity noise. In thenext two sections we describe our measurements of these two types of noisefor the Antares laser. In addition we describe the fidelity of the Antares laseras a regenerative amplifier for the injection locking laser.

Phase fidelity of the Antares laserTo verify that the injection locked Antares slave has the same spectral purityas the NPRO master, phase fidelity measurements between the master andslave lasers were made. This measurement is made by beating a fractionof the slave laser output against a fraction of the master laser output thathas been frequency shifted by 40 MHz with an acousto-optic modulator. Theheterodyne beat note is then mixed down to DC to form a phase discriminatorwhich is then analyzed using a dynamic signal analyzer.

The result of this measurement is shown in Figure 3. The measurement-phase noise is well above the sensitivity limit set by residual vibration inthe interferometric setup. At 1 KHz, the spectral noise density is 5 x 10

- 7

rad2/Hz. The RF noise spectrum can be converted into a total phase noisespectral density SO(f) using

2P,.b(f) (radian 2 2S+(f)- BP, hertz ) (2)

4

phasenolse.data

-410

1 --

*~106

10-102 \

100 1000 10000 100000frequency(Hz)

Figure 3: Phase fidelity of the regenerative slave amplifier to the masteroscillator phase by the measurement of the phase noise density SO(f).

where P8&b(f) is the single sideband power density, Pc is the carrier power,and B is the resolution bandwidth. This all optical measurement yields anupper bound for S6(f) whereas techniques relying on the closed loop errorsignal yields a lower bound. By integrating the noise spectral density overthe 100 KHz span, we estimate an upper bound added phase variance of0.3 radian from the slave Antares laser. Finally, the peak at 300 Hz is dueto vibration induced by water cooling inside the lamp pumped head. Weexpect a diode laser pumped slave laser with thermoelectric cooling to bemuch quieter. The additional phase noise corresponds to less than 1 kHzadditional linewidth.

Frequency noiseMeasurements characterizing the frequency noise spectral density of the in-jection locked Antares laser have been completed. Frequency noise measure-ments were made using the Pound/Drever [9] phase modulation discriminatorwhile the laser was locked to a high-finesse Fabry-Perot interferometer. Inthese measurements, the frequency of the laser was locked to a fundamen-tal mode of the Fabry-Perot reference cavity using a piezo to control thelaser frequency. At the low end of the frequency spectrum, well within thebandwidth of the servo, the voltage on the piezo-electric tranducer providesa good representation of the frequency noise. Above the unity gain point ofthe servo the error signal provides a measure of the free-running frequencynoise.

The Antares laser was injection locked with a 300 mW NPRO to study

5

Noise Spectral DeiSItY of InfCtIon-Seeded Angmus La

70

60

so

40

30

20

10 -

0 -

tO - I I IL|- _I_.LI I II I I I..I.II t..L.L.L .I...J.... J.J j

10 100 @000 IV

Frequency (H7)

Figure 4: Free-running frequency noise of the injection locked 18 W Antareslaser for a frequency range from 100 Hz to 10 kHz.

the frequency noise. The injection seed system includes servo control ofthe length of the slave laser (Antares) ring cavity so that it closely tracksthe frequency of the master oscillator (NPRO). At low frequencies the errorsignal of the Antares laser is nearly identical to that of the free-runningNPRO, except that there is added noise induced by the vibrations of thefan and coolant water turbulence of the Antares laser. This noise is mostlyconfined to the region from 400 Hz to 1 kHz.

At high frequencies the noise of the injection seeded Antares laser differsdramatically from the seed laser. The injection seeded, but free-running noiseis shown in Figure 4 up to 10 kHz. The most striking difference betweenthe noise of the injection seeded system and the NPRO alone is that thereare fewer noise peaks in the slave laser spectrum than appear in the masterlaser spectrum. The two most significant peaks occur at 130 kHz and 260kHz.

Overall, the noise of the injection-seeded laser closely follows the noise ofthe seed laser with some added noise due to mechanical vibrations. Never-theless, the frequency noise spectral density at 1 kHz corresponds to about50Hz/vTiz which is a factor of 50 times better than the best argon ion laser.

Intensity noiseWe have measured the intensity noise spectrum of the injection-locked Antares.Measurements were made with an InGaAs PIN photodiode followed by atransimpedance amplifier and an RF spectrum analyzer. The intensity noiseof the Antares, when injection locked to the 300 mW NPRO, has some peaks

6

I T Ell i dOn ',4 t ? 49 do@XIs .5 do/DI AL~ .5.1 do@ M93 ItFR 265R

A IjI t O %3 r,, .', dl. At 53i.33l dg. F(l St ISO e.65 lwi.

ITEN t dol -75-08 do

ED 44991 H 0-)lo de/oIV

0E 9ANWOU"H.813 h4z

CENTER 983 iN, sPA" I HMI.R9 13.3 1 i, Su , 3 l t 1@ 4.1i6 S CENTER A I@ PHE SPAN M.89 M i,

-no tio h, *V9 1.3 &1 ST 6e a ,e

Figure 5: Intensity noise of the injection locked Antares laser (left) a fre-quency range from 100 Hz to 1 MHz, and a frequency range from 100 Hz to20 MHz (right).

at integer multipies of 130 KHz as shown in Figure 5. The strongest peak isobtained at 260 KHz (70 dB above shot noise level for a resolution bandwidthof 10 KHz). These oscillations decay at higher frequency, and the intensitynoise spectrum becomes shot noise limited above 5 MHz.

6.5-W, 532-nm radiation by harmonic generationThe diode-laser-pumped solid-state NPRO has proved to be an excellentsource of stable, cw, single-axial-mode radiation for many applications in-cluding resonant cavity nonlinear frequency conversion. In an initial demon-stration of cw harmonic conversion, Kozlovsky generated 29.6 mW of 532-nm radiation from 56-mW of 1064-nm pump radiation in a MgO:LiNbO 3monolithic-external-resonant-cavity second harmonic generator.f10) The fun-damental radiation was the output of a Nd:YAG NPRO pumped by a 500-mW diode-laser array. More recently we have generated 6.5 watts of cw532-nm radiation using lithium triborate (LiB 305 or LBO) in a discrete-component external-resonant-cavity harmonic generator pumped by 18 W of1064 nm radiation.[21

36% efficient resonant harmonic generationThe setup for the externally resonant doubling experiment is shown in Figure

6. The bow-tie cavity configuration was used to reduce astigmatism; theangle of incidence on the curved mirrors is less than 3 degrees. In addition,the ring configuration eliminates feedback into the pump laser and provides

7

Figure 6: Experimental setup for resonant second harmonic generation.The flat mirrors in the bow tie resonator are separated by 40 cm, the curvedmirrors are spaced by 10.5 cm. The laser beam is incident at 3 degrees onall mirrors.

unidirectional second harmonic output. The two curved mirrors and one flatmirror in the cavity have coatings that are highly reflecting at 1064 nm withhigh transmission at 532 nm. The remaining flat mirror is the input couplerwhich has 4.2 % transmission at 1064 nm. With this layout, a tight focal spot(l/e electric field radius) of 32 microns is formed within the nonlinear crystal,located midway between the two curved mirrors. To couple the pump laserbeam into the external resonator, the external enhancement cavity resonanceneeds to be locked to the pump laser frequency. To maintain coincidence ofthe cavity resonance with the pump frequency, the cavity length is controlledwith a piezoelectric-mounted mirror through a feedback loop. The feedbackloop derives its error signal from the bea n reflected from the input couplerusing the FM .Adeband technique.J[9]

The lithium triborate LiB3 Os (LBO) nonlinear crystal used in this exper-iment was grown at the Center for Materials Research at Stanford University.The crystal was grown from high temperature solution by the top-seeded so-lution growth technique. [11] Good quality boules with diameters in excessof 30 mm have been produced. LBO is well-suited for high power secondharmonic generation (SHG) of Nd:YAG laser due to its high damage thresh-old, low absorption at both the fundamental and second harmonic, and thepossibility of type I non-critical phase-matching. The 6 mm long LBO crys-tal was heated in an oven at 149.5°C to achieve non-critical phase matching.

8

The Full Width Half Maximum (FWHM) temperature bandwidth for SHGin a 1 cm length crystal is 6.8°C.

By monitoring the leakage field through one of the high reflecting mirrorsof the doubling cavity, we deduced the circulating power inside the cavityand so determined the power enhancement factor. To maximize intracavitypower, the external cavity must be impedance matched. Measuring !osseswith and without the crystal inside the cavity, we found a loss of 0.7% dueto transmission and scattering of the three high reflecting mirrors. An addi-tional 1% loss is found for transmission through the crystal. We could notdetermine what fraction of that 1% loss is due to the anti-reflection coat-ings applied to the LBO crystal and what fraction is the result of LBO bulkabsorption and scattering lo' i. Given the 1.7% cavity round trip loss andtaking into account the additional effective loss due to conversion to the sec-ond harmonic, an available mirror with transmission of 4.2% is used as theinput coupler for best impedance matching. By carefully mode matching thepump laser into the resonant cavity using two 100 cm focal length lenses, 18watts of input power yields an intracavity circulating power of 380 watts, fora fundamental power enhancement factor of 21.

Figure .7 shows the measured and predicted second harmonic poweras a function of incident fundamental power and the corresponding con-version efficiency as a function of incident fundamental power. The solidline represents the theoretical fit calculated assuming an effective nonlinearcoefficient of 1 pm/V for LBO, and a Boyd-Kleinman focusing parameterh(B = 0, = 0.62) = 0.58. With a 6 mm long crystal, the calculated secondharmonic conversion coefficient is 6.67 x 10' per watt. In calculating thetheoretical fit, a correction taking into account the 90% transmission at 532nm of the dichroic curved mirror after the crystal has been made. With thiscorrection, the predicted and measured second harmonic power show goodagreement. The maximum second harmonic power produced is 6.5 watts with18 watts of fundamental input, representing an overall conversion efficiencyof 36%. Resonant SHG is critically dependent on losses. It is reasonable toexpect that with good mirrors and better crystal coatings that the roundtrip loss can be reduced to 0.5% and assuming perfect optical impedancematching and spatial mode matching into the external cavity, we expect asecond harmonic conversion efficiency to increase from 36% to 80%.

With almost 400 watts of circulating fundamental power and 6.5-wattsof second harmonic power, crystal heating caused by absorption at either

9

0 355 50

000

25 0 5 "LU20 0 4

15 0

> 0 °

U 0 "0

0 4 8 12 16 20Fundamental Pump (watts)

Figure 7: Resonant second harmonic generation conversion efficiency ver-sus incident fundamental power (left axis), and the corresponding secondharmonic power versus fundamental input power(right axis).

the fundamental or second harmonic wavelength becomes possible. The ef-fect of heating on single-pass second harmonic generation efficiency has beendiscussed before.[12, 13] In the single pass case, heating is evidenced by abroader, asymmetric phase matching tuning curve versus temperature, ac-companied by a shift in location of the peak. To assess the severity of crystalheating, we scanned the crystal oven temperature to measure phase match-ing tuning curve at different input pump power levels. By attenuating thefundamental input to reduce the output green power to 600 mW, we foundthe phase matching tuning curve to be symmetric with FWHM of 6.8°C,identical to the single pass case. At higher pump power levels resulting in 4watts of green output, the phase matching tuning curve skews very slightlyto the high temperature side, accompanied by less than 1 degree shift of thephase matching curve peak. From these observations, we conclude that heat-ing of LBO crystal due to absorption of the fundamental or second harmonicradiation is not significant at these powers.

Spatial mode propertiesThe spatial mode properties of radiation to be converted in a resonant non-linear optical device at high efficiency are very important. This is becausethe fundamental beam must be efficiencly mode matched into the resonantcavity and a beam with low spatial coherence cannot be efficiently coupledinto the nonlinear doubling cavity. In the next two sections we discuss the

10

7 00 10 3

30OtO-s

. ' * ' ""' p-' ' -' ,5,0010's 8 oionm .00 i0W,

500 10.3 .oi*

' ; 4.4001 04A.,®'o /zoo 10W3 0 Of

1,0010.3 .0010o-5

0.00~~~ 1& .o.dO -- i-- ,

0.40 00 0.80 1.00 1.20 1.,0 1.o 0.20 0.40 am 0.a to 1.20 1.40 1.6o .a0z, (CM) 20 (U)

Figure 8: Antares laser spatial mode. The beam profile measured witha scanning razor blade along the horizontal x-axis (left) and vertical y-axis

(right). The data are shown as dots and the fit is a gaussian profile. The

beam-waist ratio beween x and y is 1.4:1.

spatial mode properties of the Antares laser and the second harmonic fromthe doubling cavity.

Spatial mode properties of the AntaresThe transverse spatial mode structure of the slave laser determines how effi-

ciently it can be coupled into the nonlinear frequency doubling cavity used

for resonant second harmonic generation. The Antares laser output is not

diffraction limited but is astigmatic due to thermal focussing. We measured

the gaussian beam parameters of the beam along both orthogonal directionsto the beam propagation direction.

Measurements of the transverse spatial mode profile of the Antares laseroutput were made using a spinning razor blade beam chopper, a photodetec-tor and a digital oscilloscope. In Figure 8 the gaussian fit to the beam along

the two axes is presented. The beam was found to be 1.7 times diffraction

limited and astigmatic with a beam waist ratio between the x and y axes of

1.4:1.Spatial mode properties of the second harmonic

Figure 0.9 shows the measured green beam spatial profile and the calculated

Gaussian fits. As shown ir Figure 9, the beam profile along the x and y axis

are essentially Gaussian with a beam waist ratio of 1:1.1 due to non-normal

'Subsequent adjustments to the laser optical cavity have reduced this ratio to 1.05:1.

11

250 250

1 50 150100

1o.-_00Ab

€C

Sso- 0

0- 0 .0 ---. "- .4 -3 -2 -1 0 1 2 3 4 4 -3 -2 -1 0 1 2 3 4

Distance(mm) Distance(mm)

Figure 9: Measured spatial mode of the Antares laser second harmonicalong (a) x-axis and (b) y-axis. Crosses represent experimental points. Thesolid lines are Gaussian fits. The beam waist ratio between x and y axis is1:1.1.

incidence of the curved mirrors in the cavity. The beam quality is furthercharacterized by measuring the spot size after focusing by a good qualitydoublet lens and fitting the measured spot size to Gaussian propagationformula according to Siegmans M2 theory.J14] We determine the beam to beessentially diffraction limited with M2 values, representing number of timesdiffraction limited, of 1.05 and 1.01 for the x and y axes respectively. Inaddition to having good spatial beam quality, the green output has excellentfrequency stability. Spectral purity of the green output is confirmed by themeasurements of Nabors.[151 The short-time heterodyne 3dB linewidth of 15kHz demonstrates that the green preserves the phase fidelity of the pumplaser. Aside from an initial transient when the cavity is first locked to thepump, the green output power remains within 5 percent of its starting valueafter one hour of continuous operation. In addition, when the external cavitylock is lost, locking is easily re-established.

We have demonstrated efficient, high-power frequency doubling, produc-ing 6.5 watts of single frequency CW 532-nm radiation with 18-W of 1064-nmfundamental input power. The second harmonic output at 532-nm is essen-tially diffraction limited and inherits the frequency stability of the pumpsource. Such a green source is ideal as a pump for cw singly resonant para-metric oscillators.

12

13-W diode-laser-pumped Nd:YAG laserFor the last four years work has been underway to develop a 13-W, diode-laser-pumped, solid-state laser that can reach the requirements for pumpinga chirped, single-frequency, singly-resonant, parametric oscillator. This workwas begun with support from ONR and SONY corp. and has for the lasttwo years been supported by the NSF.

The most important consideration in extending diode-laser pumping ofsolid-state lasers to high cw output powers is efficient use of pump radiation.Collinear longitudinal pumping, otherwise known as end pumping, is a tech-nique in which the signal and pump beams copropagate along the length ofthe lasing medium. This geometry uses pump radiation efficiently since theoverlap between the pump and signal beams can be excellent.

End pumping suffers two limitations when extended to high power sys-tems. First, due to the small TEM0o mode area in stable resonators, a veryhigh pump brightness (in W/cm2sr - 1) is required to end pump a high powerlaser [17]. Second, and perhaps more serious, is the thermal load on the lasermedium. In any solid-state laser system, a fraction of the pump energy isdissipated as heat in the host material due to the inevitable nonradiativetransitions between the pump bands and the upper fluorescence level of thelaser. This thermal load leads to thermal gradients in the lasing region.Most solid-state lasers are fabricated in the shape of a long thin rod. Athigh pump levels this geometry suffers from thermally induced distortionssuch as thermal focusing, stress-induced biaxial focusing, and stress-inducedbirefringence. These distortions require complex correction schemes if high-efficiency, single-mode laser operation is to be achieved. The limitations ofthe rod geometry are well known and have led to studies of side-pumpedlaser systems [18].

A side-pumped laser configuration greatly reduces the pump power bright-ness requirement of the end-pumped system. Pumping through a large areainto the lasing mode is, however, less efficient than end-pumping since thepump and signal overlap is less than perfect. In a side-pumped laser, thesignal beam either passes straight through the laser medium or bounces(zigzags) off the boundaries of the medium as shown in Figure 10. Thestraight through geometry is still limited by thermal distortions in the lasermedium. The uniformly pumped and uniformly cooled zigzag slab combinesthe advantages of a rectilinear geometry for reduced stress-induced birefrin-gence and a zigzag path that eliminates thermal and stress induced focussing

13

S T

"4 L

Figure 10: The geometry of the zigzag slab laser. The dimensions of thelaser crystal are specified by the lengths L, d, the width of the slab (notshown), and the angle S. The optical path through the slab is determined bythe incidence angle D and the bounce angle T.

[191. The zigzag slab is a laser geometry capable of being scaled to arbitrarilyhigh output powers.

There is no best diode-laser-pumped, solid-state laser configuration. Thetradeoffs between simplicity and efficiency guarantee that the optimum de-sign will depend on system details such as the laser material and the physicalcharacteristics of the diode pumps. A diode pumped solid state laser ableto produce 100 W of output power in a single axial mode is currently underconstruction for DARPA in our laboratory.

Laser designThis program has focussed on the design and the construction of a cw, single-mode, 13-watt Nd:YAG diode-laser-pumped zigzag slab laser. This laser willbe pumped by 60 watts of diode laser power derived from 60 individual one-watt diode lasers as shown in figure 11. Power from the diode lasers will becoupled through optical fibers into the zigzag slab. In this subsection, we will-first consider the physical dimensions of the gain medium. This discussionwill be followed by a brief summary of the calculations performed to optimizethe output power of the design. We will then review the laser head design,the thermal modeling calculations, and the proposed optical resonator.

The physical dimensions that describe the zigzag slab are shown in Figure10. The equations that determine the slab dimensions are:

14

Diode Bank

13 W Output

Figure 11: The 13-watt diode-laser-pumped Nd:YAG oscillator. A bankof sixty external diode lasers side pumps a zigzag slab. The gain medium isshown inside a ring resonator where cylindrical lenses are used to compensatefor thermal focussing effects.

L=NdcotT, (3)

cos(s + D) = ncos(s + T), (4)

where N is the number of bounces in the slab and n is the index of refractionof the material. These equations guarantee that the bounces inside the slabare total internal reflections. Other constraints apply to specific slab config-urations such as Brewster's angle incidence or unity fill factor. Incidence atBrewster's angle guarantees low losses for one polarization mode and imposesthe following constraint on the slab dimensions

1tan(s + D) = -. ( .5)

n

We chose to manufacture a collinear (D = 0), Brewster's angle, Nd:YAG (n =1.82 at 1064nm) slab. With these constraints the slab geometry equationsreduce to: S = 28.80, T = 32.40 and Nd = 0.635L. At this point we areleft with one equation relating the two physical lengths of the slab to thenumber of bounces inside the slab. In order to minimize the scattering lossin unpumped regions of the slab, the length of the slab is chosen to be slightlygreater than the size of the pump source. With L constrained in this way,

15

we can optimize output power by varying either the thickness of the slab orthe number of bounces inside the slab.

Mathematical modeling of the output powerThe operating parameters of any laser are determined by the input-output

relation which can be expressed near threshold by

t t -)L 2 77.P .

where Po,,t is the output power of the laser, t is the output coupling, a isthe round trip loss, g is the laser gain, A1, and At are the pump and lasingwavelengths respectively, the I's are spatial overlap factors, q, is the fractionof the pump power that leads to population inversion in the laser medium,and Pin is the available pump power. These parameters are related to thephysical characteristics of a travelling wave laser by the following relations:

1=fj dxdydzso(x,y,z)ro(x,y,z), ( 7)

III

Ztt

i / 11 dxdydzs (x,y,z)ro(x,y,z), ( 8)

9 = i Pi. ( 9)

In these equations, s0 (x, y, z) represents the spatially dependent signal mode,ro(x, y, z) describes the spatially dependant pump mode, 1 is the length ofthe gain medium and Isat is the pump saturation intensity.

Output power is maximized by choosing

t = topt = Vg-a. (.10)Substituting equation 0.10 into equation 0.6 leads to the optimum input-output relation

Popt = L L2 (1 ) ni

Ao A, II1 - ,Pwhich can be analyzed in terms of four system efficiencies. The first term inEquation 11 is known as the Stokes efficiency, 'is, otherwise known as thequantum defect. This is the ultimate efficiency that any optically pumped

16

laser can achieve and represents the energy difference between the pump andsignal photons. Obviously, the designer has no control over this parameteronce the pump source and laser medium are chosen. For a diode-pumped,Nd:YAG laser pumped at 808 nm and lasing at 1064 nm, rs = 76%. Thesecond term is the coupling efficiency, 77,, and depends on the pumping ge-ometry chosen. End pumped lasers can have coupling efficiencies exceeding80%, while 40% is typical for side pumped lasers. The third term is the laserefficiency, which is dominated by losses in the laser cavity. For example, atypical low gain laser with a 20% single pass gain has a laser efficiency of60% at 1% loss, and 47% at 2% loss. Low loss is the most important factorleading to high-efficiency, diode-pumped lasers. The last term is the absorp-tion efficiency, which depends mostly on the size of the gain medium relativeto the absorption length of the pump. Designing a laser is an exercise inoptimization. One must balance laser and absorption efficiencies, which typ-ically improve with larger gain media, against coupling efficiency, which hasthe opposite tendency, all the while keeping the constraint of low losses inmind.

As we saw above, the slab geometry constrains us to vary either the thick-ness of the slab, or the number of bounces inside the slab. For convenience,we chose to model output power versus N. The final laser parameter thatmust be chosen is the TEM~o mode radius of the resonator cavity. Althoughwe are free to choose this radius, there are two factors to consider: the fillingfactor and the aperture losses. A large beam radius fills the volume of theslab well but may be severely clipped by the slab edges. Losses are the majorobstacle to efficient laser performance, so we choose a beam radius of 300 Pmthat suffers very little clipping while still filling a large fraction of the slab.The largest TEM00 mode radius practically attainable in a stable resonatoris about 700 pm.

Figure 12 shows results of numerical calculations that relate couplingand absorption efficiency versus the number of bounces. We see that absorp-tion efficiency and coupling efficiency are approximately inverses. Couplingefficiency is approximately linearly related to bounce number, indicating thatthe signal beam spends more time in the strongly pumped region. Absorp-tion efficiency decreases as the bounce number increases since the slab thinswith increasing bounce number. These effects combine to produce a flat out-put power versus N curve, which decreases at high bounce numbers due toclipping losses at the input aperture. We chose a 10 bounce slab for low

17

90%I i1

80%...... ...-

70% . ......... ................................. .. .• - 60% ............. ... ............ ............. . ................ ... ....... ..

50% .............. ................................ ... .....,.. .....,. . ."- " -- : "-60%

SO% . .. . •...

............ t........ ......... ............... .......... ..30% ............. .- ............. ................ ....... . ._ _20% . .. .... ..

10%4 6 8 10 12 14 16 is 20

Number of bounces

Figure 12: Variations in absorption efficiency and coupling efficiency withnumber of bounces. Absorption efficiency decreases due to the decreasingthickness of the slab. Coupling efficiency increases as the signal beam spendsmore of its time in the strongly pumped regions of the slab. The product ofthese two efficiencies is approximately constant with bounce number.

loss operation and to avoid a very thin design that would be difficult tomanufacture. Figure 13 shows the final dimensions of the zigzag slab.

Head designOnce the dimensions of the Nd:YAG slab have been established a laser headmust be designed. The laser head must perform the following functions:it must support the crystal, it must allow easy access for the pump light,it must not obstruct the signal beam, and it must allow for heat removalfrom the laser crystal. Traditional flashlamp-pumped rod lasers are cooledby water flowing around the rod, this system is very good at removing heatbut suffers from a number of problems. The most important of these is thevibrations introduced onto the laser rod by the turbulent cooling water flow.These vibrations lead to laser frequency noise and should be avoided in orderto simplify the servo requirements of the frequency stabilization system.

One of the most important advantages of diode pumping is that the signif-icantly reduced heat load on the laser medium greatly simplifies the coolingsystem required. Our approach is to use thermal conduction to cool thecrystal. A preliminary head design is shown in Figure .14. In this head weuse the cold finger approach to cool the sides of the laser crystal, therebyallowing easy pump access. A thermoelectric cooler pulls the heat out of theassembly. Calculations show that for 25-W of heat deposited in the laser

18

1.5mm --.- '.-- ..- %- - Il24 mm r*-

1.5 mm

Figure 13: Final dimensions of the zigzag slab laser. The final design callsfor a 1.5 x 1.5 x 24mm' Nd:YAG slab. The optical beam will bounce off theslab walls ten times.

TEC Glass Spacer

Nd:YAG

Optical

Axis

Cu Cold Finglers

Cooling Water

Figure 14: Preliminary laser head design. Copper cold fingers are used tocool the slab. A thermoelectric cooler pulls heat out of the assembly anddeposits it in the heat sink.

19

crystal, a cooler current of 1.8 A is required to cool the surface of the gainmedium to room temperature, assuming a heat sink temperature of 25°C.The heat sink is required to conduct 32.5-W of heat away from the laserhead.

Thermal analysisWe have modeled the 13-W zig-zag slab laser using several computer pro-grams in order to calculate the single-pass distortions of a wavefront propa-gating through the slab. The slab is modeled as a finite element grid with8960 elements and 11277 nodes. The heat loading in the slab due to thepumping of the laser diodes is modeled using a Monte-Carlo ray tracingtechnique. Using experimentally measured values for the Nd:YAG absorp-tion and the beam parameters of the fiber-coupled diode light, the programtraces pump rays as they bounce through the slab and then computes theheat deposited in each element in the grid. These data are then used asinput into a finite element analysis program which solves the steady stateheat equation for the temperature distribution in the slab. A third programuses ray tracing techniques to calculate the phase distortion introduced intothe laser beam due to the bulk change in the index of refraction of Nd:YAGwith temperature.

We have predicted that the wavefront distortion due to the bulk temper-ature effects can be accurately modeled by a cylindrical lens. For example,for a heat loading of 15-W in a 1.5xl.5x26 mm 3 slab, the thermal distor-tions is very nearly a perfect cylindrical lens of focal length 40 cm in thedirection of heat removal.. This thermal lensing must be taken into accountwhen modeling the optical resonator.

We have also performed calculations on the thermally induced stress, asshown in Figure 15. The stress contours on the surface of the slab nearlyfollow the pumping geometry, i.e., the position of the fibers on the slab.Finally, this modelling has shown that the heat loading for this laser is lessthan the stress fracture limit of Nd:YAG, so that thermal management andhead cooling issues are straightforward to engineer.

Optical resonatorThe proposed optical resonator is a very simple, stable, four mirror designbased on the resonator used in the injection-locked Antares laser system. No-table differences are the lack of an intracavity polarizer, since the Brewster'sangle faces on the slab define the laser polarization. Stability is achievedby employing cylindrical lenses and the thermal focussing described above.

20

Figure 15: Surface stress resulting from the thermal loading of the 13-watt,slab laser. The bright areas show large stresses near the pump source. Thecalculated stresses are far below the fracture limit for the material.

Mode sizes on the order of 300 pm are expected in the gain medium.Once we have built the laser we will investigate and compare its operation

to the design theory by measuring the output power of the slab laser as afunction of diode-laser-pump power, output coupling, and optical resonatordesign. In addition, we will measure the depolarization, the spectral densityof noise, and the beam wiggle of the laser output as a function of pumppower. Unidirectional oscillation and frequency stability will be achievedthrough injection locking rather than with intracavity etalons to keep cavitylosses at a minimum and efficiency and output power high.

Diode laser pumpingAn important component of any laser is the pump source. The zigzag slablaser is pumped by a bank of sixty SLU-304XRs: one-watt, fiber-coupled,diode lasers manufactured by Sony Corp. Besides the aforementioned advan-tages of diode pumping, fiber coupling allows the laser designer to separatethe problems of cooling the laser crystal and the pump diodes. Also, theuse of fiber connectors allows replacement of failed diodes while the laser isoperating. These advantages come at the expense of pump brightness. Effi-cient (70%) coupling of a diode laser's high aspect ratio rectangular emitterto a circular core fiber typically reduces the source brightness by an order ofmagnitude, with a similar reduction in laser gain. This low gain laser regimedemands low loss designs for efficient operation.

21

Diode laser driversDiode lasers require a highly regulated constant current source to drive themand a temperature control system to tune their output wavelength to thenarrow absorption bands of Nd:YAG. In this subsection we describe the de-sign and performance of the system used to power and cool the 1-W, diodelasers.

Our approach has been to purchase the diode laser drivers and man-ufacture the temperature controllers. We purchased an LDS 10000 seriesrack mounted laser diode driver from Light Control Instruments of San LuisObispo, CA. These diode drivers are based on the LCI 500 series regulatedcurrent sources. Our drivers consist of two crates, each with thirty channels.Each channel is capable of driving a maximum of 4 A with a complianceof 5 V, with less than 0.05% rms noise and greater than 100 ppm/0 C tem-perature stability. The current sources are protected by an uninterruptablepower supply system to insure against premature diode failure due to spikesor blackouts in the building power.

Temperature ControlCommercially-available, high-power diode lasers are typically 20 to 40% effi-cient, with an emission linewidth of 2 to 5 nm. This implies that a thermalmanagement system must be applied to remove roughly two watts of heatfor every watt of optical power generated by the diode, while maintaininga jun,.ion temperature near 25°C. The junction temperature needs to beaccurately controlled since the emission wavelength of diode lasers vary by0.3 nm/°C. Given that the strong 808 nm absorption band of Nd:YAG hasa linewidth of approximately 2 nm, the junction temperature needs to becontrolled to better than PC of the setpoint. The SLU-304XR laser packageincludes a 10k thermistor for temperature measurement and a thermoelectriccooler for thermal management. We have calculated that the thermoelectriccooler current required to cool all sixty diode lasers to an emission wave-length of 808 nm is as follows: an average current of 790 mA per diode, anda total current of 48 A, at a diode case temperature of 5C. This reducedcase temperature greatly relaxes the current driver requirements but requiresmounting the diodes on cold plates, cooled by a recirculating water chiller.

The requirements on the sixty channel temperature control system are asfollows: a maximum output current of 2 A per channel, a thermistor tem-perature sensor input, and ±0.1*C accuracy, for negligible laser amplitudenoise contributions due to pump power absorption fluctuations. Figure 0.16

22

R2 ¢ CR SepointRI

Figure 16: The design can be divided into three stages. The error stage

produces a voltage proportional to temperature error. The control stage

provides frequency compensation for low residual error and optimal timeresponse. The power stage provides the drive necessary to cool the diode

lasers.

shows a simple diagram of the design. The system consists of three separately

optimized stages: the error stage, the control stage, and the power stage.As shown in Figure 16, the error signal is generated by employing a

balanced bridge. The precision resistor divider is used to establish a reference

which is compared to the difference in resistance between the thermistor andthe setpoint potentiometer. The INAI01 instrumentation amplifier produces

an output that has a slope of 1 V/0C for small deviations from zero error.A warning signal is generated if the error exceeds 5 V, indicating possible

damage to the diode laser.The control stage is simply a single opamp implementing a P1 control

law. Even with the very long time constants associated with a large thermal

mass. this stage can generate large dc gain and near optimal damping ratios.

The power stage consists of a single LH01O1 high current amplifier, capable

of delivering 5 A continuous output current while operating from a total

supply voltage of 10 V. This second property is crucial in reducing power

consumption.Figure 17 shows the error signal of the temperature control unit, under

actual operating conditions. The trace shows that, the system achieved adiode junction temperature within i0 mC of the setpoint with peak fluctua-tions less than 5 mC from the mean. The sixty temperature control channels

23

2 3.2770004 1.QEE0 1 2 3 4 S 6 7 a

Elapsed time (min)

Figure 17: Temperature controller error vs time. The diode temperatureis kept within 15mC of the setpoint under actual operating conditions.

are housed in three separate CAMAC crates, which provide the necessary dcpower and a method of interfacing to a remote monitoring computer.

Experimental resultsWe have designed and built an end-pumped laser to test the laser diodesand temperature controllers in a realistic application. This laser system alsoprovides a good way to test the predictions of the theory presented above.As shown in Figure 18, the laser consists of a single pump laser, collimationand focussing optics, a 3mm diameter by 20 mm long rod of 1.1% Nd:YAG,and an output coupler with a 1 m radius of curvature. One end of the laserrod is coated to be highly reflecting at 1064 nm and highly transmitting at810 nm, while the other end is coated to be highly transmitting at 1064 nm.

The spatial beam parameters of the pump and signal beams were mea-sured using a scanning razor blade chopper, and analyzed using the M2 theoryof non-diffraction limited optical beams [14]. These measurements show abeam quality factor of 100 for the fiber output beam. This large M2 is due tothe high numerical aperture fiber required for efficient ,'iode laser coupling..Typical beam qualities for high power diode laser beams are 2 in the planeperpendicular to the diode junction and 25 in the plane parallel to the diodejunction. The high M2 values of the fiber coupled diodes dramatically showthe brightness reduction that accompanies fiber coupling.

The last laser parameter that needs to be determined is the round triploss. The theory predicts the following relationship between threshold powerand output coupling

24

Cu Heat Sink

Coupling Output CouplerOptics Nd:YAG Rod

Plane Mirror AR 1.06 I~m

HR 1.06 ILMHT 810 nm

Sony, IlW

O Laser

Figure 18: A schematic of the end-pumped rod laser experiment. Theoutput of one optical fiber is focussed into the end of a coated Nd:YAG rod.

A standing-wave resonator is established by the curved output coupler.

5%

0%L0.10 015 0.3 ol 0.30 03n

Threho d power (W)

Figure 19: Findlay-Clay analysis of the rod laser. The intercept of the plotof output coupling versus threshold power yields the loss in the laser cavity,here 1.5%.

25

2

0,

ISO ......... .. ..... .. .. ...........

10

0 100 200 M0 4M5 M M M MInput Power (MW)

Figure 20: Input-output power graph for the rod laser. The open circlesare the measured output power and the solid line is the prediction of themodelling program. The only free parameter in the calculation is used tomatch the predicted threshold power to that observed. The, laser achieves180mW of output power for 700 mW of pump power.

a + t E -qaPth, (0.12)

therefore the intercept of a plot of output coupling versus threshold poweryields the round trip loss in the laser. This procedure is known as the Findlay-Clay analysis [39]. The results for our laser are shown in Figure 19.

With these experimental parameters we can numerically integrate theinput-output equation to predict the laser output power. The results ofsuch a calculation are compared with experimental data in Figure 20. Thepredicted threshold power was close to that observed, but the theory under-estimates the slope efficiency by 10%. This shows the strength of the theoryin predicting the performance of diode-laser-pumped lasers.

Pulsed optical parametric oscillatorsThe potential of optical parametric oscillators (OPO's) for converting laseroutput to tunable radiation at arbitrarily selected frequencies was recognizedwith the first OPO demonstration.(201 The early parametric oscillators weredoubly resonant. Doubly resonant oscillators (DRO's) have an advantage oflow pumping threshold for oscillation, but DRO's also have complex tun-ing properties and stable DRO operation is difficult. These problems were

26

demonstrated in early experiments and analysis [20] and became more ap-parent with the first demonstration of continuously pumped OPO's.[21, 22]Further investigation demonstrated the stringent cavity tolerances requiredfor stable DRO operation.[23] Theoretical analysis shows that when thesetolerances are satisfied, the DRO will provide many useful and unique prop-erties for generating tunable radiation with coherence that nearly reproducesthe pump coherence.

Singly resonant optical parametric oscillators (SRO's) have less compli-cated tuning properties than DRO's, and SRO's are free of some of the con-straints that make stable operation of DRO's difficult. Pump thresholds,however, are higher for SRO's, and it has been difficult demonstrating cwoperation in these oscillators. The double resonance condition also providesfrequency selectivity that easily results in single-mode-pair oscillation. Thisfrequency selection is not available with single resonance unless frequencyselecting components are added to the SRO.

SRO's have been forced to operate single-mode on the resonated signalwave even with highly multimode pumping.[24] The combined frequency se-lection of phase matching, a dispersing grating, and an intracavity etalonwere necessary to achieve single-mode oscillation with these conditions. Themulti-axial-mode pump resulted in a multimode output at the nonresonatedidler field. When this type of frequency control is used with a pulsed SROpumped with an injection-seeded Q-switched laser with single-mode output,the result is single-mode output at both the signal and idler wavelengths.[25]With the addition of piezoelectric control of the cavity length and computercontrol of all the adjustable parameters, spectrographic measurements with300-MHz resolution were possible.[26] These SRO's used angle tuned LiNbO3pumped at 1.06 pm.

SRO pumped by a Q-switched laserThe single-mode pump alone is not sufficient to produce single-mode-pairoscillation in a simple SRO with no frequency selection except that of phasematching. This was observed in a BaB20 4 SRO pumped with the 354-nmthird harmonic of the output of an injection seeded Q-switched Nd:YAGlaser.[28] The single frequency pump did reduce OPO output energy fluctu-ation to 10% from 30% which was observed with multimode pumping wheninjection seeding of the Q-switched laser was blocked. The plane-parallel-cavity SRO with 1.2-cm-long BaB20 4 crystal and 3-cm overall length pumpedby a 6-nsec pulse would oscillate on typically 8 axial modes at the signal

27

wave. The broad tuning range of the BaB2 O4 0PO extending from 412nm to 2.55 pm included both the 1.064-pm fundamental and 532-nm secondharmonic of the laser, which were available for injection seeding the SRO.With injection seeding at 532 nm or 1.064 pm, the buildup time of the SROoscillation was substantially decreased, and single-mode-pair oscillated wasachieved. Measurements of the SRO threshold yielded values that were lessthan predicted from calculations based on reported values of the nonlinearcoefficient of BaB2 0 4 . This observation lead to a series of measurementsthat resulted in a reassessment of the scale for optical nonlinear coefficients.Injection seeding is an effective means of obtaining single-mode operationof and SRO.[29, 30, 31] A stable cw DRO would be an excellent radiationsource for this application.

Long-pulse-pumped SRO'sGreater frequency selection is possible with longer buildup times of paramet-ric oscillation. Pump pulses of 500-nsec duration were derived from a diode-pumped NPRO by gating the cw output followed by multipass amplificationin a flashlamp pumped laser amplifier.[32] A monolithic SRO was pumpedby the second harmonic generated by the oscillator-amplifier laser system.Pumping with the 532-nm harmonic allowed noncritical phase matching inthe 5%MgO:LiNbO 3 crystal. The purpose of this experiment was to investi-gate spectral narrowing and to reduce OPO threshold to a level approachingthat which could be achieved directly by diode-pumped lasers.

The monolithic SRO tuned from 834 to 958 nm and 1.47 to 1.2 pm whentemperature was adjusted between 1900 and 125°C.[3] Damage limitation ofthe MgO:LiNbO 3 SHG crystal required that the 5-kW output of the laseramplifier be no longer than 500 ns. Under these conditions 800 W of 532-nmharmonic was generated. The ring-cavity configuration of the SRO allowedhigh efficiency; up to 60% pump depletion was observed after threshold wasreached. About 20% of the time single-mode operation of the SRO wasobserved. More often, however, simultaneous oscillation on three axial modeswas observed, and occasionally as many as 8 modes oscillated simultaneously.Thus long-buildup time alone was not sufficient to guarantee single-modeoscillation.

Pulse-pumped doubly resonant optical parametric oscillation Afurther reduction in OPO threshold is obtained with the doubly resonant os-cillator (DRO) configuration in which the OPO is resonant at both the signaland idler wavelengths. The added constraint of double resonance in addition

28

to phase matching and conservation of energy, however, makes stable op-eration of the DRO difficult.[31 The frequency stability of the NPRO andthe mechanical stability of monolithic DRO construction are useful in over-coming this difficulty. A doubly resonant monolithic DRO was constructedfrom MgO:LiNbO 3 with broad-band dielectric mirrors highly reflecting near1.06 prm coated on the crystal.[33] A ring geometry was formed by usingreflections from two 10-mm radius-of curvature surfaces on the ends of thenoncritically phase-matched, 1.25-cm-long crystal and a polished flat on oneside for total internal reflection (Figure 21). The DRO was pumped at 532nm by second-harmonic pulses generated from the relaxation oscillations of aNd:YAG NPRO. Pulsed operation was required because the DRO thresholdwas marginally higher than could be produced by the NPRO and harmonicgenerator when operated cw.

A 10% modulation of the diode-laser current at 325 kHz drove the Nd:YAGNPRO into relaxation oscillations. The 1.06-p m fundamental pulses had 260-mW peak power and were efficiently converted into 400-ns, 230-mW, 532-nmpulses by externally resonant second-harmonic generation. The buildup ofparametric oscillation occupied most of the pump pulse duration. OverallDRO efficiency was only 7% due to the long buildup time. After thresholdwas reached 60% pump depletion occurred. The DRO tuned between 1.02and 1.12 pm by adjustment of both temperature and electric field. The mostremarkable aspect of the DRO operation was that single-mode-pair oscilla-tion was achieved on almost every pulse. The only exception was when theDRO was tuned between cluster centers and either simultaneous or alternat-ing output on widely spaced modes approximately 4 nm apart were observed.The monolithic DRO with a pulsed single-mode pump will oscillate on a sin-gle mode pair due to the constraint of double cavity resonance.

CW optical parametric oscillatorsOur cw OPO experiments have used the second harmonic radiation de-scribed above for pumping. The availability of 532-nm pump radiation andnoncritical phase matching with a reasonable temperature tuning range inMgO:LiNbO 3 was important for these experiments. Temperature-tuned non-critically phase-matched monolithic resonators provided mechanical stabilityand low loss for these DRO studies. A drawing of a ring-path monolithic OPOis shown in Figure 21. Pulsed radiation of 400-ns duration and cw 532-nmradiation produced by the monolithic resonant cavity harmonic generator was

29

-12.5 mm ' 2.2 m

T- 1o'C

Figure 21: Monolithic DRO geometry used for experimental observations

used to pump the DRO's.[34] The DRO performance was remarkable. Thethreshold for the cw device was only 11 mW. Single-mode-pair output wasobserved routinely in the cw DRO's. As much as 80% pump depletion wasobserved at two times above threshold.The output coupling of the OPO wasnot optimized for maximum output. Nevertheless, the slope efficiency shownin Figure 22 was 64% and surprisingly linear. The free-running cw DROwithout servo control would oscillate for periods of one minute on a singlemode pair. At degeneracy the DRO would stably produce the subharmonicof the pump for periods of 20 minutes.

The coherence properties of the cw DRO are also remarkable.[15] It wasobserved during the periods of stable operation between mode hops that thecoherence of the outputs of the DRO reproduced the coherence of the pumpradiation. When operated on a mode pair adjacent to degeneracy, the signaland the idler difference frequency was stable to better than 1 kHz, the limitof resolution of the measurement. The difference frequency of the signal andthe idler is an indication of the additional frequency noise that the DROadds to that present on the pump radiation. At degeneracy, the output ofthe DRO was a phase-locked subharmonic of the pump radiation. This wasshown by interference of the DRO output and the laser radiation used togenerate the second harmonic that was in turn used to pump the DRO.

Our measurements of DRO performance and coherence provided the ba-sis for a theoretical analysis of the frequency-tuning and -control propertiesof these devices.435] The analysis showed how three tuning parameters arerequired for controlling two cavity resonances and phase matching in the

30

10.

Slope efficiency =6%

0

0 5 10 15 20 25

Pump power (moW)Figure 22: Output power as a function of pump power for a cw monolithicMgO:LiNbO 3 DRO. The DRO was pumped at 532 nm and the output wasnear degeneracy at 1.064 pm.

DRO. It was possible to model the observed tuning properties of mono-

lithic MgO:LiNbO 3 DRO's using temperature dependent dispersion, thermalexpansion, electro-optic, and piezoelectric properties of the material. Theanalysis is applicable to both monolithic and discrete component cavities.The tuning analysis can be used to calculate the tolerances for stable DROoperation. These tolerances are stringent, for example 0.001°C temperature

stability and 5 x 1010 m cavity length stability are typical requirements.More stringent tolerances than these, however, are achieved in laser frequencystabilization. Lasers are now available that provide the required frequencystability for resonant cavity nonlinear frequency conversion techniques. It is

an interesting and important challenge to reproduce the available laser fre-

quency stability in the output of a nonlinear frequency conversion device.

31

Publications resulting from this programThe publications resulting from the Office of Naval Research contract ONRN00014-88-K-0701, "An All-Solid-State Chirped Source for Coherent OpticalRadar". Reprints of these publications are contained in the Appendix.

W. J. Kozlovsky, E. K. Gustafson, R. C. Eckardt and R. L. Byer, "Ef-ficient monoolithic MgO:LiNbO 3 singly resonant optical parametric oscilla-tor," Opt. Lett. 13, 1102 (1988).

R. L. Byer, "Frequency stable high power lasers in space," RelativisticGravitational Experiments in Space, edited by R. W. Hellings (NASA Head-quarters, Washington, D.C, 1989), pp. 166-170.

M. K. Reed and R. L. Byer, "Mode-locked operation af a Nd:YLF laserand amplification in a Q-switched Nd:glass slab laser," IEEE J. QuantumElectron. 26, pp. 1399-1404 (1990).

C. D. Nabors and R. L. Byer, "Monolithic Optical Parametric Oscillatorsfor Quantum Optics," Coherence and Quantum Optics VI, edited by J. H.Eberly, L. Mandel and E. Wolf (Plenum, New York, 1990), pp. 787-791.

C. D. Nabors and R. M. Shelby, "Two-color squeezing and sub-shot-noisesignal recovery in doubly resonant optical parametric oscillators," Phys. Rev.42, pp. 556-559 (1990).

C. D. Nabors, S. T. Yang, T. Day and R. L. Byer, "Coherence propertiesof a doubly-resonant monolithic optical parametric oscillator," J. Opt. Soc.Am. B 7, pp. 815-820 (1990).

M. K.Reed and R. L. Byer, "The output beam quality of a Q-switchedNd:Glass slab laser," IEEE J. Quantum Electron. 26, pp. 2138-2145 (1990).

R. C. Eckardt, C. D. Nabors, W. J. Kozlovsky and R. L. Byer, "Opticalparametric oscillator frequency tuning and control," J. Opt. Soc. Am B 8,pp. 646-667 (1991).

R. C. Eckardt, H. Masuda, Y. X. Fan and R. L. Byer, "Absolute and rela-

32

tive nonlinear optical coefficients of KDP, KD*P, BaB204, Li1O3, MgO:LiNbO3,and KTP measured by phase-matched second harmonic generation," IEEEJ. Quantum Electron. 26, pp. 922-933 (1990).

T. Day, E. K. Gustafson and R. L. Byer, "Active frequency stabilizationof diode laser pumped, nonplanar ring oscillators," Proc. SPIE 1223, pp.181-185 (1990).

T. Day, A. D. Farinas and R. L. Byer, "Demonstration of a low bandwidth1.06 m optical phase-locked loop for coherent homodyne communication,"IEEE Photonics Tech. Lett. 2, pp. 294-296 (1990).

S.T. Yang, C.C. Pohalski, E.K. Gustafson, R.L. Byer, R.S. Feigelson,R.J. Raymaker, and R.K. Route,"6.5-W, 532-nm radiation by cw resonantexternal-cavity second-harmonic generation of an 18-W Nd:YAG laser inLiB 3O5 ," Optics Letters, Vol. 16, No. 19, October 1, 1991, 1493-1495.

R. L. Byer, "Advances in nonlinear optical materials and devices," to bepublished in Proceedings of the 5th Toyota Conference on Nonlinear OpticalMaterials, Aichi-ken, Japan, 6-9 October 1991, S. Miyata, ed. (Elsevier Sci-ence, Amsterdam, 1992).

T. Day, E. K. Gustafson and R. L. Byer, "Sub-Hertz frequency stabi-lization of two diode-laser pumped-Nd:YAG lasers locked to a Fabry-Perotinterferometer," submitted to J. of Quantum Electronics.

R. C. Eckardt, T. Day, E. K. Gustafson and R. L. Byer, "Frequency stableoperation of Nd:YAG Lasers," to be published, Proceedings of the 5th In-ternational School on Laser Applications in Atomic, Molecular and NuclearPhysics, (Vilnius University Press, Vilnius, Lithuania).

33

Dissertations resulting from this programThe dissertations resulting from the Office of Naval Research contract ONRN00014-88-K-0701, "An All-Solid-State Chirped Source for Coherent OpticalRadar".

W.J. Kozlovsky, "Efficient nonlinear conversion of frequency-stable lasers,"Department of Applied Physics, Stanford University, November 1988.

C.D. Nabors, "Coherence and two-color squeezing in doubly resonantparametric oscillators," Department of Applied Physics, Stanford Univer-sity, December 1989.

A.C. Nilsson, "Eigenpolarization theory and experimental linewidth studyof monolithic nonplanar ring oscillators," Ph.D. dissertation, Dept. of Ap-plied Physics, Stanford University, Stanford California, 1989.

M. K. Reed, "Nd:Glass slab laser development and applications for x-ray lithography," Department of Applied Physics, Stanford University, June1990.

T. Day, "Frequency stabilized solid state lasers for coherent optical com-munications," Dept. of Electrical Engineering, Stanford University, StanfordCalifornia, 1990.

R.C. Eckardt, "Continuous tuning and frequency stabilization of dou-bly resonant optical parametric oscillators," Department of Applied Physics,Stanford University, June 1991

34

Personnel supported by this programThe personnel supported by the Office of Naval Research contract ONRN00014-88-K-0701, "An All-Solid-State Chirped Source for Coherent OpticalRadar".

Professor Robert L. ByerProfessor Martin M. Fejer

Senior Research Associate Robert EckardtResearch Associate Eric Gustafson

Graduate Student Alex FarinasGraduate Student Steven Yang

Graduate Student Chris PohalskiGraduate Student Stephan SchillerGraduate Student David NaborsGraduate Student Murray Read

Graduate Student William J. Kozlovsky

35

ConclusionWe have reported on progress under the Office of Naval Research contractONR N00014-88-K-0701, "An All-Solid-State Chirped Source for CoherentOptical Radar". This progress includes seven principle accomplishments in-cluding: the demonstration of an 18-watt, single-frequency, injection-locked,TEM00, Nd:YAG laser, the measurement of the phase fidelity of this re-generative amplifier and the frequency, and intensity noise of the injectionlocked laser, the conversion of the output of this laser to 6.5 watts of single-frequency, TEM0o, green radiation at 532 nm by resonant second harmonicgeneration, the measurement of the spatial mode characteristics of this lasersand its second harmonic, the progress toward a 13-watt, single-frequency,diode-laser-pumped, solid-state laser, the construction and testing of a pulsedparametric oscillator, and the construction of a cw doubly resonant paramet-ric oscillator and the use of this 80% efficient device for the generation ofsqueezed states.

36

References

[1] C.D. Nabors, A.D. Farinas, T. Day, S.T. Yang, E.K. Gustafson and R.L.Byer,"Injection locking of a 13-W cw Nd:YAG ring laser", Opt. Lett.14, pp. 1189-1191 (1989).

[2] S.T. Yang, C.C. Pohalski, E.K. Gustafson, R.L. Byer, R.S. Feigelson,R.J. Raymaker, and R.K. Route,"6.5-W, 532-nm radiation by cw res-onant external-cavity second-harmonic generation of an 18-W Nd:YAGlaser in LiB 3 O5 ", Optics Letters, Vol. 16, No. 19, October 1, 1991, 1493-

1495.

[3] W. J. Kozlovsky, E. K. Gustafson, R. C. Eckardt and R. L. Byer, "Effi-cient monolithic MgO:LiNbO 3 singly resonant optical parametric oscil-lator," Opt. Lett. 13, 1102 (1988).

[4] C.D. Nabors, R.M. Shelby,"Two-color squeezing and sub-shot-noise sig-nal recovery in doubly resonant optical parametric oscillators," Phys.Rev. A, 42, 556-559 (1990).

[5] T. Day, E.K. Gustafson and R.L. Byer,"Sub-Hertz frequency stabiliza-tion of two diode laser pumped Nd:YAG lasers locked to a Fabry-Perotinterferometer," To be published in the special issue of the Journal ofQuantum Electronics in March 1992.

[6] R. Adler, Proc. IRE 34, 351 (1946).

[7] T.J. Kane and R.L. Byer, "Monolithic unidirectional single-modeNd:YAG ring laser", Optics Letters, vol. 10, pp. 65-67, (1985).

37

[8] T.J. Kane, A.C. Nilsson, and R.L. Byer, "Frequency stability and off-set locking of a laser-diode-pumped Nd:YAG monolithic nonplanar ringoscillator," Opt. Lett., 12, 175-177 (1987).

[9j R.W.P. Drever, J.L. Hall, F.V. Kowalski, J. Hough, G.M. Ford, A.J.Munley, and H. Ward, "Laser Phase and Frequency Stabilization usingan Optical Resonator," Appl. Phys. B 31, 97-105 (1983).

[101 W. J. Kozlovsky, C. D. Nabors and R. L. Byer, IEEE J. QuantumElectron. 24, 913 (1988).

[111] R.S. Feigelson and R.K. Route, Proc. Eighth American Conference onCrystal Growth, pp. 123b, Vail, Colorado, July 1990.

[12] M. Okada, S. leiri, IEEE J. Quantum Electron. QE-7, pp. 469-470(1971).

[131 M. Okada, S. Ieiri, IEEE J. Quantum Electron. QE-7, pp. 560-563(1971).

[141 Anthony E. Siegman, Lasers, Ch. 11, (Mill Valley: University ScienceBooks, 1986).

[15] C.D. Nabors, S.T. Yang, T. Day and R.L. Byer, "Coherence propertiesof a doubly resonant monolithic optical parametric oscillator," J. Opt.Soc. Am. B, Vol 7, No. 5, May 1990, pp. 815-820.

[16] K. Kubodera and K. Otsuka, "Single-Transverse-Mode LiNbP 4O12 SlabWaveguide Laser," J. Appl. Phys 50 (2) Feb. 1979

[171 T.Y. Fan and A. Sanchez, "Pump source requirements for end-pumpedlasers," IEEE Journal of Quantum Electronics, Vol. 26, No. 2, February1990, pp.3 1 1-3 16 .

[18] J.M. Eggelston, T.J. Kane, K. Kuhn, J. Unternahrer, and R.L. Byer,"The slab geometry laser-Part I: Theory," IEEE J. Quant. Electron.QE-20, 289 (1984).

[19] W.S. Martin and J.P. Chernoch, "Multiple Internal Reflection FacePumped Laser," US Patent 3633 126, 1972.

38

[20] J. A. Giordmaine and R. C. Miller, in Physics of Quantum Electronics,P.L. Kelly, B. Lax and P. E. Tannenwald, eds. (McGraw, New York, 1965),pp. 31-42.

[211 R. G. Smith, J. E. Geusic, H. J. Levinstein, J. J. Rubin, S. Singh andL. G. Van Uitert, Appl. Phys. Lett. 12, 308 (1968).

[221 R. L. Byer, M. K. Oshman, J. F. Young and S. E. Harris, Appl. Phys.Lett. 13, 109 (1968).

[23] R. G. Smith, IEEE J. Quantum Electron. QE-9, 530 (1973).

[24] S. J. Brosnan and R. L. Byer, IEEE J. Quantum Electron. QE-15, 415(1979).

[25] T. K. Minton, S. A. Reid, H. L. Kim and J. D. McDonald, Opt. Commun.69, 289 (1989).

[26] T. K. Minton, H. L. Kim, S. A. Reid and J. D. McDonald, J. Chem.Phys. 89, 6550 (1988).

[27] Y. X. Fan, R. C. Eckardt, R. L. Byer, J. Nolting and R. Wallenstein,Appl. Phys. Lett. 53, 2014 (1988).

[281 T.Y. Fan and R.L. Byer, "Diode-laser-pumped solid-state lasers," IEEEJ. Quant. Electron. QE-24, 6 (1988)

[29] J. E. Bjorkholm and H. G. Danielmeyer, Appl. Phys. Lett. 15, 171(1969).

[30] V. L. Boichenko, M. M. Novikov and A. I. Kholodnykh, Soy. J. QuantumElectron. 17, 392 (1987).

•[31] J. G. Haub, M. J. Johnson, B. J. Orr and R. Wallenstein, Appl. Phys.Lett. 58, 1718 (1991).

[32] T. J. Kane, W. J. Kozlovsky and R. L. Byer, Opt. Lett. 11, 216 (1986).

[331 W. J. Kozlovsky, C. D. Nabors, R. C. Eckardt and R. L. Byer, Opt.Lett. 14, 66 (1989).

39

[34] C. D. Nabors, R. C. Eckardt, W. J. Kozlovsky and R. L. Byer, Opt.Lett. 14, 1134 (1989).

[35] R. C. Eckardt, C. D. Nabors, W. J. Kozlovsky and R. L. Byer, J. Opt.Soc. Am. B 8, 646 (1991).

[36] R.L. Byer,"Diode laser-pumped solid-state lasers," Science, 239,742-747(1988).

[37] W.J. Kozlovsky, C.D. Nabors and R.L. Byer,"Efficient second harmonicgeneration of a diode-laser-pumped cw Nd:YAG laser using monolithicMgO:LiNbO 3 external resonant cavities," IEEE J. Quantum Electron.QE-24, pp. 913-919 (1988).

[38] D.H. Jundt, M.M. Fejer, R.L. Byer, R.G. Norwood, and P.F. Bordui,"69% Efficient continuous-wave second-harmonic generation in lithium-rich lithium niobate," Opt. Lett, Vol. 16, No. 23, December 1, 1991,pp.1856-1858.

[39] D. Findlay and R.A. Clay,"The measurement of internal losses in 4-level lasers," Physics Letters, Volume 20, number 3, 15 February 1966,pp.277-278.

[40] A.C. Nilsson, "Eigenpolarization theory and experimental linewidthstudy of monolithic nonplanar ring oscillators," Ph.D. dissertation,Dept. of Applied Physics, Stanford University, Stanford California, 1989.

[41] A.C. Nilsson, E.K. Gustafson, and R.L. Byer, "Eigenpolarization theoryof monolithic nonplanar ring oscillators," J. Quantum Elec., 25, April1989, pp. 76 7-7 9 0 .

[42] W.J. Kozlovsky, C.D. Nabors and R.L. Byer, "Second-harmonic genera-tion of a continuous-wave diode-pumped Nd:YAG laser using an externalresonant cavity," Opt Lett 12, 12 (1987), pg. 1014

[43] W.J. Kozlovsky "Efficient Resonant Harmonic Generation of a Nd:YAGLaser," Dissertation, Stanford University December 1988

[44] W.J. Kozlovsky, C.D. Nabors, R.C. Eckardt, and R.L. Byer,"MonolithicMgO:LiNbO 3 doubly resonant optical parametric oscillator pumped by

40

a frequency-doubled diode-laser-pumped Nd:YAG laser," Opt. Lett., 14,66-68 (1989).

[45] W.J. Kozlovsky, C.D. Nabors, and R.L. Byer,"Efficient second harmonicgeneration of a diode-laser-pumped CW Nd:YAG laser using monolithicMgO:LiNbO3 external resonant cavities," IEEE J. Quantum Electron.,Vol. 24, No. 6, June (1988) pp. 913-919.

[46] C.D. Nabors, W.J. Kozlovsky, and R.L. Byer, "Efficient second har-monic generation of a diode-pumped cw Nd:YAG laser using an exter-nally resonant cavity," SPIE (1988).

[47] M.K. Reed, R.L. Byer,"Performance of a conduction cooled Nd:Glassslab laser," SPIE 736 (1987).

[48] M.K. Reed, W.J. Kozlovsky, R.L. Byer, G.L. Harnagel and P.S. Cross,"Diode Laser Array Pumped Nd:YAG and Nd:Glass Laser Source," postdeadline paper, Topical Meeting on Coherent Laser Radar: Technologyand Applications, Aspen, Colorado, July 27-31, 1987.

[49] M.K. Reed, W.J. Kozlovsky , R.L. Byer, G.L. Harnagel, and P.S.Cross, "Diode Laser Array Pumped Nd:YAG and Nd:Glass Slab Os-cillators,"Opt. Lett., 13, 204-206, March, 1988.

41